Fig. 3

Statistical and partitioning analysis of MPM patient data. Panel I. Distribution of overall survival vs TMN by gender. Distribution of OS showed no statistically significant difference if compared to TNM disease stage, except for a very slight difference related to stage IB MPM in females (A) vs males (B). Positive values show pairs of means that are significantly different. Standard deviation error bars are shown as well. Panel II. Partition analysis for overall survival of whole data. The split criterion divides the tree through a top-down clustering algorithm. Results are represented in female (C) and male (D) subgroups, respectively. Count: number of training observation; G2: Gini index. Lower values indicate better fit. Panel III. Comparison of progression free survival vs chemotherapy schedule. Distribution of PFS showed no statistically significant difference if compared to first line chemotherapy in both females (E) vs males (F). Positive values show pairs of means that are significantly different. Standard deviation error bars are shown as well. Panel IV. Comparison of overall survival vs second line chemotherapy schedule. Distribution of OS showed moderate significant difference in female patients treated with vinorelbine vs those treated with gemcitabine (G); no statistically significant differences can be shown in males (H). Positive values show pairs of means that are significantly different. Standard deviation error bars are shown as well. OS: overall survival, PFS: progression free survival, R: first line re-challenge chemotherapy, V: vinorelbine, G: gemcitabine, P: platinum. P-A: platinum-pemetrexed, NO: second line chemotherapy not performed